U.S. patent number 7,664,596 [Application Number 11/427,728] was granted by the patent office on 2010-02-16 for air traffic demand prediction.
This patent grant is currently assigned to Lockheed Martin Corporation. Invention is credited to Daniel Joseph Cleary, Bradley A. Culbertson, Jonathan Dehn, Louis John Hoebel, John Michael Lizzi, Paul W. Mettus, Liviu Nedelescu, Rajesh Venkat Subbu, Gerald Bowden Wise.
United States Patent |
7,664,596 |
Wise , et al. |
February 16, 2010 |
**Please see images for:
( Certificate of Correction ) ** |
Air traffic demand prediction
Abstract
Systems and methods for airspace demand prediction with improved
sector level demand prediction are provided. In one embodiment, an
air traffic demand prediction system (10) operable to predict
demand within an airspace divided into sectors includes an expanded
route predictor (14) operable to generate predicted two-dimensional
expanded route information (40) associated with at least one
requested flight (34), a trajectory modeler (16) operable to
generate predicted four-dimensional expanded route information
(46), a sector crossing predictor (18) operable to generate
predicted sector crossing information (48), a departure time
predictor (22) operable to generate predicted departure time
information (54), and a demand modeler (62) operable to generate a
demand model (28), the demand model (28) including predicted time
intervals associated with the at least one requested flight
indicating when it is expected to be present within one or more
sectors of the airspace.
Inventors: |
Wise; Gerald Bowden (Clifton
Park, NY), Lizzi; John Michael (Wilton, NY), Hoebel;
Louis John (Burnt Hills, NY), Subbu; Rajesh Venkat
(Clifton Park, NY), Cleary; Daniel Joseph (Schenectady,
NY), Nedelescu; Liviu (Arlington, VA), Mettus; Paul
W. (Gaithersburg, MD), Culbertson; Bradley A.
(Gaithersburg, MD), Dehn; Jonathan (Damascus, MD) |
Assignee: |
Lockheed Martin Corporation
(Bethesda, MD)
|
Family
ID: |
38877730 |
Appl.
No.: |
11/427,728 |
Filed: |
June 29, 2006 |
Prior Publication Data
|
|
|
|
Document
Identifier |
Publication Date |
|
US 20080004792 A1 |
Jan 3, 2008 |
|
Current U.S.
Class: |
701/120; 701/533;
701/465 |
Current CPC
Class: |
G08G
5/0043 (20130101) |
Current International
Class: |
G06F
17/00 (20060101); G06F 19/00 (20060101) |
References Cited
[Referenced By]
U.S. Patent Documents
Other References
Wang et al.; Modeling time and space metering of flights in the
national airspace system; Proc. of the 2004 Winter Simulation
Conf.; IEEE. cited by examiner .
Fan et al.; Cluster model for flight demand forecasting; Proc. of
the 5th World Congress on Intelligent Control and Automation; Jun.
15-19, 2004; IEEE: p. 3170-3173. cited by examiner .
Oussedik et al.; Dynamic air traffic planning by genetic
algorithms; Proc. on the 1999 Congress on Evolutionary Computation;
Jul. 6-9, 1999; IEEE, vol. 2, pp. 1110-1117. cited by examiner
.
Fellman et al.; Estimating efficacy of progressive planning for air
traffic flow management; Proc. of the 2004 Winter Simulation Conf.;
IEEE. cited by examiner.
|
Primary Examiner: Zanelli; Michael J.
Attorney, Agent or Firm: Marsh Fischmann & Breyfogle
LLP
Claims
What is claimed is:
1. An air traffic demand prediction system operable to predict
demand within an airspace divided into sectors, said system
comprising: an expanded route predictor, the expanded route
predictor being operable to generate predicted two-dimensional
expanded route information associated with at least one requested
flight, the at least one requested flight having an associated
departure location and an associated destination location; a
trajectory modeler receiving the predicted two-dimensional expanded
route information, the trajectory modeler being operable to
generate predicted four-dimensional expanded route information
associated with the at least one requested flight; a sector
crossing predictor receiving the predicted four-dimensional
expanded route information, the sector crossing predictor being
operable to generate predicted sector crossing information
associated with the at least one requested flight, the sector
crossing information including times when the at least one
requested flight is expected to cross from one sector of the
airspace into another sector of the airspace; a departure time
predictor, the departure time predictor being operable to generate
predicted departure time information associated with the at least
one requested flight; and a demand modeler operable to generate a
demand model, the demand model including predicted time intervals
associated with the at least one requested flight indicating when
the at least one requested flight is expected to be present within
one or more sectors of the airspace, the predicted time intervals
being derived from at least the predicted sector crossing
information and the predicted departure time information.
2. The system of claim 1 wherein the predicted two-dimensional
expanded route information includes geographic position fixes
defining a route expected to be flown by the at least one requested
flight between its departure location and its destination
location.
3. The system of claim 1 wherein the predicted four-dimensional
expanded route information includes geographic position fixes
defining a route expected to be flown by the at least one requested
flight between its departure location and its destination location,
altitudes associated with the geographic position fixes, and times
associated with the geographic position fixes.
4. The system of claim 1 wherein the expanded route predictor
receives historical data including information relating to
previously completed instances of one or more flights corresponding
with the at least one requested flight, geometric cluster data
derived from information relating to previously completed flights
between the same departure location and destination location as
those associated with the at least one requested flight, and flight
information parameters associated with the at least one requested
flight.
5. The system of claim 4 further comprising: a schedule retriever,
the schedule retriever being operable to retrieve a flight
schedule, wherein the flight schedule includes the flight
information parameters relating to the at least one requested
flight.
6. The system of claim 1 wherein the trajectory modeler further
receives anticipated cruise speed and cruise altitude information
associated with the at least one requested flight.
7. The system of claim 1 further comprising: an enroute traffic
retriever operable to retrieve enroute information corresponding to
enroute flights associated with the requested flight, the enroute
information being input to the trajectory modeler.
8. The system of claim 1 wherein the departure time predictor
receives historical departure delay information including
information relating to previously completed instances one or more
flights corresponding with the at least one requested flight.
9. The system of claim 1 further comprising: a demand model
interface, the demand model interface being operable to present the
demand model to a user of the air traffic demand system.
10. The system of claim 9 wherein the demand model interface
comprises a graphical user interface displayable on a display
device.
11. The system of claim 1 further comprising: a response filter
receiving the predicted sector crossing information from the sector
crossing predictor, the response filter being operable to filter
the predicted sector crossing information to obtain filtered
predicted sector crossing information, the filtered predicted
sector crossing information being used by the demand modeler with
the predicted departure time information to derive the predicted
time intervals.
12. The system of claim 11 wherein the demand modeler includes: a
graph generator receiving the filtered predicted sector crossing
information and the predicted departure time information, the graph
generator being operable to generate a temporal constraint graph
corresponding with each sector of the airspace entered or exited by
the at least one requested flight along a route expected to be
flown by the at least one requested flight between its departure
location and its destination location, each temporal constraint
graph being derived from the predicted sector crossing information
and the predicted departure time information and representing
predicted time intervals associated with the at least one requested
flight indicating when the at least one requested flight is
expected to be within the sector of the airspace corresponding with
the graph.
13. The system of claim 1 wherein said expanded route predictor,
said trajectory modeler, said sector crossing generator, said
departure time predictor, and said demand modeler comprise
instructions executable by one or more processors.
14. A method of predicting air traffic demand within an airspace
divided into sectors, said method comprising the steps of:
performing an expanded route prediction for at least one requested
flight within the airspace, the at least one requested flight
having an associated departure location and an associated
destination location; performing a departure prediction for the at
least one requested flight; performing a temporal congestion
prediction for the at least one requested flight using results of
the expanded route prediction; and generating a demand model based
on results of the temporal congestion prediction and the departure
prediction, the demand model including predicted time intervals
associated with the at least one requested flight indicating when
the at least one requested flight is expected to be present within
one or more sectors of the airspace entered or exited on its route
from its associated departure location to its associated
destination location.
15. The method of claim 14 wherein said step of performing an
expanded route prediction comprises: retrieving flight information
parameters associated with the at least one requested flight;
retrieving historical data including information relating to
previously completed instances of one or more flights corresponding
with the at least one requested flight; retrieving geometric
cluster data derived from information relating to previously
completed flights between the same departure location and
destination location as those associated with the at least one
requested flight; and generating predicted two-dimensional expanded
route information including geographic position fixes defining a
route expected to be flown by the at least one requested
flight.
16. The method of claim 15 further comprising: utilizing a flight
schedule including the flight information parameters relating to at
least one requested flight.
17. The method of claim 14 wherein said step of performing a
temporal congestion prediction comprises: receiving predicted
two-dimensional expanded route information; generating predicted
four-dimensional expanded route information including geographic
position fixes defining a route expected to be flown by the at
least one requested flight between its departure location and its
destination location, altitudes associated with the geographic
position fixes, and times associated with the geographic position
fixes; and generating predicted sector crossing information
including times when the at least one requested flight is expected
to cross from one sector of the airspace into another sector of the
airspace.
18. The method of claim 17 wherein said step of performing a
temporal congestion prediction further comprises: receiving enroute
information associated with the at least one requested flight; and
using the enroute information to obtain four-dimensional expanded
route information associated with each enroute flight associated
with the at least one requested flight.
19. The method of claim 17 wherein said step of performing a
temporal congestion prediction further comprises: receiving
anticipated cruise speed and cruise altitude information associated
with the at least one requested flight.
20. The method of claim 14 wherein said step of performing a
departure time prediction comprises: retrieving flight information
parameters associated with the at least one requested flight;
querying historical departure delay information to identify
previously completed instances of one or more flights having flight
information parameters similar to the flight information parameters
of the at least one requested flight; and generating a delay
distribution for the at least one requested flight based on the
identified previously completed instances of one or more
flights.
21. The method of claim 14 further comprising: outputting the
demand model to one or more individuals responsible for directing
air traffic within the airspace.
22. The method of claim 21 wherein said step of outputting
comprises displaying parameters of the demand model in a graphical
user interface on a display device.
23. The method of claim 14 wherein said step of generating a demand
model comprises: generating a temporal constraint graph
corresponding with each sector of the airspace entered or exited by
the at least one requested flight along a route expected to be
flown by the at least one requested flight between its departure
location and its destination location, each temporal constraint
graph being derived from the predicted sector crossing information
and the predicted departure time information and representing
predicted time intervals associated with the at least one requested
flight indicating when the at least one requested flight is
expected to be within the sector of the airspace corresponding with
the graph.
24. The method of claim 14 further comprising: filtering the
results of the temporal congestion prediction prior to said step of
generating a demand model, wherein in said step of generating a
demand model the predicted time intervals are derived from the
results of the departure prediction and the filtered results of the
temporal congestion prediction.
25. A system for predicting air traffic demand within an airspace
divided into sectors, said system comprising: an expanded route
predictor that performs an expanded route prediction for at least
one requested flight within the airspace, the at least one
requested flight having an associated departure location and an
associated destination location; a departure time predictor that
performs a departure prediction for the at least one requested
flight within the airspace; a temporal congestion predictor that
performs a temporal congestion prediction for the at least one
requested flight within the airspace using results of the expanded
route prediction; and a demand modeler that generates a demand
model based on results of the temporal congestion prediction and
the departure prediction, the demand model including predicted time
intervals associated with the at least one requested flight
indicating when the at least one requested flight is expected to be
present within one or more sectors of the airspace.
26. The system of claim 25 wherein said expanded route predictor
comprises instructions executable by one or more processors to
generate predicted two-dimensional expanded route information
associated with the at least one requested flight.
27. The system of claim 25 wherein said expanded route predictor
generates predicted two-dimensional expanded route information
associated with the at least one requested flight, and wherein said
temporal congestion predictor comprises instructions executable by
one or more processors to generate predicted four-dimensional
expanded route information associated with the at least one
requested flight using at least the predicted two-dimensional
expanded route information and to generate predicted sector
crossing information associated with the at least one requested
flight using at least the predicted four-dimensional expanded route
information.
28. The system of claim 25 wherein said departure time predictor
comprises instructions executable by one or more processors to
generate predicted departure time information associated with the
at least one requested flight.
29. The system of claim 25 wherein said temporal congestion
predictor generates predicted sector crossing information
associated with the at least one requested flight, wherein said
departure time predictor generates predicted departure time
information associated with the at least one requested flight, and
wherein said demand modeler comprises instructions executable by
one or more processors to generate the demand model from at least
the predicted sector crossing information and the predicted
departure time information.
30. The system of claim 25 further comprising: a graphical user
interface displayable on a display device, wherein the graphical
user interface presents the demand model to a user of the system.
Description
FIELD OF THE INVENTION
The present invention relates generally to air traffic control, and
more particularly to predicting airspace demands.
BACKGROUND OF THE INVENTION
The aviation community faces increasing flight delays, security
concerns and airline costs. Industry stakeholders such as the
Federal Aviation Administration (FAA), the airlines, and the
Transportation Security Agency operate in a complex real-time
environment with layered dependencies that make the outcome of air
traffic management initiatives hard to predict. Thus, planning of
air traffic initiatives in more detail, and further in advance,
such that the national airspace system can be managed more
efficiently has become increasingly important. One key requirement
for enacting an air traffic system with higher emphasis on
strategic management of traffic is accurately predicting air
traffic demand within various airspaces.
Controlled airspaces are typically subdivided into a number of
sectors, and generally an individual air traffic controller is
responsible for controlling air traffic within a particular sector.
The number of flights expected to be in a particular sector during
a time period of interest is the demand for that sector. Since one
air traffic controller can reasonably be expected to monitor and
direct only a limited number of flights (e.g., 10 to 15 flights) at
the same time within the sector for which they are responsible, it
is desirable to determine the expected demand within sectors of a
controlled airspace and the effect that an individual flight
request will have on the expected demand at some time in the future
so that the flights within an airspace can be directed
appropriately to help keep the anticipated number of flights within
the sectors of the airspace within manageable levels. A limited
number of systems and methods are currently applied to the problem
of air traffic demand predictions. One example of such a system is
the FAA's enhanced air traffic management system (ETMS). However,
many of these methods and systems are not sufficiently accurate,
particularly under non-standard environments, such as convective
weather situations, in order to effectively predict air traffic
demand.
SUMMARY OF THE INVENTION
Accordingly the present invention provides systems and methods for
airspace demand prediction with improved sector level demand
prediction enabling air traffic controllers to achieve smoother and
more expeditious flow of air traffic. In this regard, improved
sector level air traffic demand predictions are achieved through
the use of advantageous techniques such as flight path clustering,
case based route selection, and prediction of departure and sector
crossing times using temporal reasoning techniques. Through use of
such advanced techniques, an increase in accuracy over existing
systems performing similar air traffic demand prediction functions
is obtained. For example, by employing geometric clustering
techniques to a larger set of historical data, air traffic demand
predictions made in accordance with the present invention can be
more accurate, and by employing temporal prediction techniques,
such as temporal reasoning, a probabilistic approach to air traffic
demand prediction is utilized.
In one aspect of the invention, an air traffic demand prediction
system includes an expanded route predictor, a trajectory modeler,
a sector crossing predictor, a departure time predictor, and a
demand modeler. The air traffic demand prediction system operates
to predict demand within an airspace divided into sectors.
The expanded route predictor operates to generate predicted
two-dimensional expanded route information associated with one or
more requested flights. Each requested flight has an associated
departure location and an associated destination location. The
destination and departure locations are typically airports,
although they may be airstrips, landing pads, or other fixed or
movable locations from which airplanes, helicopters, airships and
other flying vehicles may take off and land. The predicted
two-dimensional expanded route information may include geographic
position fixes defining a route expected to be flown by each
requested flight between its associated departure location and its
destination location.
In order to generate the expanded route information, the expanded
route predictor may receive historical data including information
relating to previously completed instances of one or more flights
corresponding with the requested flight(s), geometric cluster data
derived from information relating to previously completed flights
between the same departure location(s) and destination location(s)
as those associated with the requested flight(s), and flight
information parameters associated with the requested flight(s). In
this regard, the air traffic demand prediction system may include a
schedule retriever that operates to retrieve a flight schedule
including the flight information parameters relating to the
requested flight(s).
The trajectory modeler receives the predicted two-dimensional
expanded route information and operates to generate predicted
four-dimensional expanded route information associated with the
requested flight(s). In this regard, the predicted four-dimensional
expanded route information may include geographic position fixes
defining a route expected to be flown by the each requested flight
between its departure location and its destination location,
altitudes associated with the geographic position fixes, and times
associated with the geographic position fixes. In addition to
receiving the predicted two-dimensional expanded route information,
the trajectory modeler may also receive anticipated cruise speed
and cruise altitude information associated with the requested
flight(s).
The sector crossing predictor receives the predicted
four-dimensional expanded route information and operates to
generate predicted sector crossing information associated with the
requested flight(s). The predicted sector crossing information
includes times when the requested flight(s) is/are expected to
cross from one sector of the airspace into another sector of the
airspace.
The air traffic demand prediction system may also include a
response filter. The response filter receives the predicted sector
crossing information from the sector crossing predictor and
operates to filter the predicted sector crossing information to
obtain filtered predicted sector crossing information. The filtered
predicted sector crossing information may be used by the demand
modeler with predicted departure time information to derive
predicted time intervals.
The departure time predictor operates to generate predicted
departure time information associated with the requested flight(s).
In this regard, the departure time predictor may receive historical
departure delay information from which the predicted departure time
information may be derived. The historical departure delay
information may include information relating to previously
completed instances of one or more flights corresponding with the
requested flight(s).
The demand modeler operates to generate a demand model. In this
regard, the demand model includes predicted time intervals
associated with the requested flight(s) indicating when the
requested flight(s) is/are expected to be present within one or
more sectors of the airspace. The demand modeler derives the
predicted time intervals from at least the predicted sector
crossing information (or from the filtered predicted sector
crossing information when a response filter is included in the air
traffic demand prediction system) and the predicted departure time
information.
To facilitate use of the information included in the demand model,
the air traffic demand prediction system may further include a
demand model interface. The demand model interface operates to
present the demand model to a user (e.g., an air traffic
controller) of the air traffic demand system for utilization
thereby and interaction therewith. In this regard, the demand model
interface may comprise a graphical user interface displayable on a
display device.
In one embodiment, the demand modeler comprises a graph generator.
The graph generator receives the predicted sector crossing
information and the predicted departure time information and
operates to generate a temporal constraint graph corresponding with
each sector of the airspace entered or exited by each requested
flight along an associated route expected to be flown by each
requested flight between its departure location and its destination
location. Each temporal constraint graph is derived from the
predicted sector crossing information and the predicted departure
time information and represents predicted time intervals associated
with each requested flight indicating when each requested flight is
expected to be within the sector of the airspace corresponding with
the graph.
The air traffic demand prediction system may include an enroute
traffic retriever. The enroute traffic retriever receives enroute
data associated with the requested flight(s) and operates to
provide updated enroute information associated with the requested
flight(s) using the enroute data. The updated enroute information
is input to the trajectory modeler to obtain four-dimensional
expanded route information corresponding to the associated enroute
data.
In another aspect of the invention, a method of predicting air
traffic demand within an airspace divided into sectors includes
performing an expanded route prediction for one or more requested
flights within the airspace, performing a temporal congestion
prediction for the requested flight(s) using results of the
expanded route prediction, performing a departure prediction for
the requested flight(s), and generating a demand model based on
results of the temporal congestion prediction and the departure
prediction. Each requested flight has an associated departure
location and an associated destination location, and the
destination and departure locations may, for example, be airports,
airstrips, landing pads, or other fixed or movable locations from
which airplanes, helicopters, airships, and other flying vehicles
may take off and land. The demand model that is generated includes
predicted time intervals associated with each requested flight
indicating when each requested flight is expected to be present
within one or more sectors of the airspace entered or exited on its
route from its associated departure location to its associated
destination location.
The step of performing an expanded route prediction may include
retrieving flight information parameters associated with the
requested flight(s), retrieving historical data including
information relating to previously completed instances of one or
more flights corresponding with the requested flight(s), retrieving
geometric cluster data derived from information relating to
previously completed flights between the same departure location
and destination location as those associated with the requested
flight(s), and generating predicted two-dimensional expanded route
information including geographic position fixes defining a route
expected to be flown by each requested flight. In this regard, the
method may further include the step of utilizing a flight schedule
including the flight information parameters relating to the
requested flight(s).
The step of performing a temporal congestion prediction may include
receiving the predicted two-dimensional expanded route information,
generating predicted four-dimensional expanded route information,
and generating predicted sector crossing information including
times when the requested flight(s) is/are expected to cross from
one sector of the airspace into another sector of the airspace. The
four-dimensional expanded route information may include geographic
position fixes defining a route expected to be flown by each
requested flight between its departure location and its destination
location, altitudes associated with the geographic position fixes,
and times associated with the geographic position fixes. The step
of performing a temporal congestion prediction may further include
receiving updated enroute information associated with the requested
flight(s) and using the updated enroute information to obtain
four-dimensional expanded route information associated with the
enroute information. The step of performing a temporal congestion
prediction may also further include receiving anticipated cruise
speed and cruise altitude information associated with the requested
flight(s) that is used together with the other received information
in generating the predicted four-dimensional expanded route
information and generating the predicted sector crossing
information.
The step of performing a departure time prediction may include
retrieving flight information parameters associated with the
requested flight(s), querying historical departure delay
information to identify previously completed instances of one or
more flights having flight information parameters similar to the
flight information parameters of the requested flight(s), and
generating a delay distribution for each requested flight based on
the identified previously completed instances of one or more
flights.
The step of generating a demand model may include generating a
temporal constraint graph corresponding with each sector of the
airspace entered or exited by each requested flight along an
associated route expected to be flown by each requested flight
between its departure location and its destination location. In
this regard, each temporal constraint graph is derived from the
predicted sector crossing information and the predicted departure
time information and represents predicted time intervals associated
with each requested flight indicating when each requested flight is
expected to be within the sector of the airspace corresponding with
the graph.
The method of predicting air traffic demand may also include
filtering the results of the temporal congestion prediction prior
to the step of generating a demand model. In this regard, in the
step of generating a demand model, the predicted time intervals may
be derived from the results of the departure prediction and the
filtered results of the temporal congestion prediction.
The method of predicting air traffic demand may further include
outputting the demand model to one or more individuals responsible
for directing air traffic within the airspace. In this regard, the
step of outputting may include displaying parameters of the demand
model in a graphical user interface on a display device.
These and other aspects and advantages of the present invention
will be apparent upon review of the following Detailed Description
when taken in conjunction with the accompanying figures.
DESCRIPTION OF THE DRAWINGS
For a more complete understanding of the present invention and
further advantages thereof, reference is now made to the following
Detailed Description, taken in conjunction with the drawings, in
which:
FIG. 1 is a block diagram of one embodiment of an air traffic
demand prediction system;
FIG. 2 is a diagrammatic view of one embodiment of a case-based
retrieval process;
FIG. 3 is a plot depicting clustering of exemplary completed flight
routes from San Francisco to Chicago O'Hare;
FIG. 4 is a diagrammatic view of one embodiment of a departure
delay prediction process;
FIG. 5A is a diagrammatic view of one embodiment of a graph
generation process;
FIG. 5B is plot showing an exemplary temporal constraint graph;
FIG. 6 depicts an exemplary solution obtained by the graph
generation process; and
FIG. 7 depicts one embodiment of a graphical user interface of a
demand model interface of the air traffic demand prediction
system.
DETAILED DESCRIPTION
FIG. 1 shows one embodiment of an air traffic demand prediction
system 10. The air traffic demand prediction system 10 analyzes one
or more requested flights to determine the effect of the requested
flight(s) on the demand within various sectors of a controlled
airspace during a time period of interest.
The air traffic demand prediction system 10 includes a schedule
retrieval component 12, an expanded route prediction component 14,
a trajectory modeling component 16, a sector crossing component 18,
an enroute traffic retrieval component 20, a departure time
prediction component 22, a response filter component 24, and a
graph generation component 26. Such components 12-26 may also be
referred to herein as the schedule retriever 12, the expanded route
predictor, the trajectory modeler, the sector crossing predictor
18, the enroute traffic retriever 20, the departure time predictor
22, the response filter 24, and the graph generator 26. In the
present embodiment, the various components 12-26 of the air traffic
demand prediction system 10 are implemented in software
instructions executable by one or more processors. In other
embodiments, one or more of the components 12-26 of the air traffic
demand prediction system may be implemented in hardware or in
programmable logic (e.g., in a field programmable gate array)
instead of software.
Using various inputs, the components 12-26 of the air traffic
demand prediction system 10 generate a demand model 28. The demand
model 28 is provided to a demand model interface 30 for
presentation to and utilization by a user of the air traffic demand
system 10. In this regard, the demand model interface 30 may be a
graphical user interface (GUI) displayable on a display device such
as, for example, a computer monitor. In this regard, the demand
model interface 30 may be implemented in software instructions
executable by one or more processors. In other embodiments, the
demand model interface 30 may be a non-graphical interface and it
may be implemented in hardware or in programmable logic (e.g., in a
field programmable gate array) instead of software.
The schedule retrieval component 12 operates to retrieve a flight
schedule 32. The schedule retrieval component 12 may retrieve the
flight schedule 32 by combining various sources of information
including published flight schedules (e.g., the official airline
guide (OAG)) available from various airlines and air charter
services. The flight schedule 32 includes flight information
relating to one or more flights scheduled to depart during a time
period of interest. In this regard, the flight information in the
flight schedule 32 may include, for example, airline, aircraft
type, scheduled departure time, departure airport and destination
airport for each flight in the schedule 32. The time period of
interest may, in general, be a block of time of any desired length
starting at any time in the future. However, in one embodiment, the
time period of interest is a one-hour period commencing fifteen
hours in the future. The duration of the time period of interest
and/or when such time period commences may be fixed in the air
traffic management system 10 or variable based on, for example,
user selected preferences during start-up of the air traffic demand
prediction system 10 and/or user input during operation of the
system 10.
Once the flight schedule 32 is created for a time period of
interest, a flight request 34 may be selected from the flight
schedule 32 for subsequent processing by the air traffic demand
prediction system 10. The flight request 34 may also be referred to
herein as the requested flight 34. The flight information from the
flight schedule 32 for the requested flight 34 is input to the
expanded route prediction component 14. Additionally, further
information 58 relating to the requested flight 32 may be input to
the trajectory modeling component 16. Of particular significance to
the trajectory modeling component 16 is the cruise speed and cruise
altitude of the flight request 32. Such additional information
(e.g., cruise speed, cruise altitude) 58 may be associated with
flights included in the schedule 32 by the schedule retrieval
component 12 from historical data and/or predictive algorithms.
The expanded route prediction component 14 receives as inputs the
flight information for the flight request 34 and also geometric
cluster data 36 relating to air traffic routes and historical data
38 relating to air traffic routes. The historical data 38 includes
information describing individual flight paths taken by completed
flights from departure airports to destination airports. Such
information may comprise geographic position fixes specified by,
for example, latitude and longitude (lat/long points) associated
with the various segments of an individual flight path. The
geometric cluster data 36 includes averages or other combinations
of the information describing similar individual flight paths taken
by completed flights from departure airports to destination
airports. In this regard, the geometric cluster data 36 may be
obtained from the historical data 38 as described in connection
with FIG. 3.
All of the historical data 38 and the geometric cluster data 36
available may not necessarily be relevant to the particular flight
request 34 being processed since the historical data 38 and the
geometric cluster data 36 available may relate to completed flights
between different departure and/or destination airports than those
in the flight information associated with the flight request 34. In
this regard, only historical data 38 and geometric cluster data 36
associated with flights between the same departure and destination
airports as in the flight information associated with the flight
request 34 being processed may be selected from the historical data
38 and the geometric cluster data 36 for input to the expanded
route prediction component 14. For example, if the requested flight
34 originates in San Francisco and is destined for Chicago O'Hare,
then historical data 38 and geometric cluster data 36 relating to
completed flights from San Francisco to Chicago O'Hare may be
selected as the relevant data for input to the expanded route
prediction component 14.
Using flight information for the flight request 34, relevant
cluster data 36 and relevant historical data 38 as inputs, the
expanded route prediction component 14 operates to generate
predicted two-dimensional expanded route information (predicted
ER.sub.2d) 40 associated with the flight request 34. In this
regard, the predicted ER.sub.2d 40 associated with the flight
request 34 includes predicted geographic position fixes (e.g.,
lat/long points) that define a route expected to be flown by the
requested flight 34 from its departure airport to its destination
airport. Such predicted route will involve one or more, and often
many, air traffic control sectors within the airspace from the
departure airport to the destination airport.
The enroute traffic retrieval component 20 operates to generate a
set of zero or more enroute flights associated with the flight
request 34 for input to the trajectory modeling component 16. An
enroute flight consists of two-dimensional expanded route
information along with cruise speed and cruise altitude
(collectively enroute information 42). In this regard, the enroute
information 42 may be obtained from a database of enroute data 44.
The enroute data 44 may, for example, include information from a
flight plan filed for the requested flight 34 prior to departure
and/or actual information transmitted from the flight and/or
obtained by systems monitoring the airspace traversed by the
flight.
The trajectory modeling component 16 receives the predicted
ER.sub.2d 40 from the expanded route prediction component 14 along
with the additional flight information 58 (e.g., predicted cruise
speed and cruise altitude) associated with the requested flight 34.
Using these inputs, the trajectory modeling component 16 operates
to generate predicted four-dimensional expanded route information
(predicted ER.sub.4d) 46. In this regard, the predicted ER.sub.4d
46 includes geographic position fixes (e.g., latitude/longitude
points) that define a route expected to be flown by the requested
flight 34 from its departure airport to its destination airport
along with altitude and times associated with such geographic
position fixes. Also, when available, the enroute information 42
from the enroute traffic retrieval component 20 is input to the
trajectory modeling component 16 to provide an enhanced picture of
airspace demand in addition to the airspace demands imposed by the
requested flight 34 being processed.
The sector crossing component 18 receives the predicted ER.sub.4d
46 from the trajectory modeling component 16. Using the predicted
ER.sub.4d 46 as an input, the sector crossing component 18 outputs
predicted sector crossing information 48 to the response filter
component 24. In this regard, the predicted sector crossing
information 48 includes predicted four-dimensional entry and exit
points (e.g., latitude, ongitude, altitude, and time) for the
airspace sectors along the predicted route of the requested flight
34.
As shown, the trajectory modeling component 16 and the sector
crossing component 18 may be part of another air traffic control
related system 60. One example of a suitable system 60 is the
Lockheed Martin User Request Evaluation Tool (LM URET) system 60.
Such a system 60 has been installed in Air Route Traffic Control
Centers (ARTCCs) and includes trajectory modeling and sector
crossing components 16, 18 suitable for interfacing with or
incorporating into the air traffic demand prediction system 10. In
other embodiments, the trajectory modeling component 16 and/or the
sector crossing component 18 may be components that are only
included within the air traffic demand prediction system 10.
The response filter component 24 receives the predicted sector
crossing information 48 from the sector crossing component 18. The
response filter component 24 operates to filter the predicted
sector crossing information 48 to obtain filtered predicted sector
crossing information 50. In this regard, the filtered predicted
sector crossing component filters the predicted sector crossing
information 48 to format times and durations into a standard format
and to remove duplicate or otherwise unnecessary sector crossing
information.
Using historical departure delay time data 52 as an input, the
departure time prediction component 22 generates predicted
departure time information 54 for the requested flight 34. In this
regard, the predicted departure time information 54 may include a
temporal interval during which the requested flight is predicted to
depart. A departure time prediction process that may be utilized by
the departure time prediction component 22 to generate the
predicted departure time information 54 is described in connection
with FIG. 4.
The filtered predicted sector crossing information 50 and the
predicted departure time information 54 are input to the graph
generation component 26. Using these inputs, the graph generation
component 26 generates a temporal constraint graph 56 representing
predicted time intervals for various segments of the requested
flight 34 (e.g., predicted early, middle and late entry times into
and exit times from various sectors to be traversed by the
requested flight 34) along its predicted route.
In one embodiment, the temporal constraint graph 56 generated for
each segment of the predicted route may be a Tachyon graph. Tachyon
is a computer software implementation of a constraint-based model
for representing and reasoning about qualitative and quantitative
aspects of time. The Tachyon software may also be referred to
herein as the Tachyon temporal reasoner. The Tachyon temporal
reasoner was developed by General Electric Global Research Center
(GE GRC). In other embodiments, software and/or hardware providing
sufficiently similar functionality may be employed in place of the
Tachyon temporal reasoner. An exemplary Tachyon graph 56 is
depicted and described in connection with FIG. 5B.
The graph generation component 26 and the Tachyon graph(s) 56
generated thereby may comprise a demand model generation component
62. In other embodiments, the demand model generation component 62
may include additional elements. The output from the demand model
generation component 62 (e.g., graph(s) 56) is used to update the
demand model 28 that is provided to the demand model interface 30
for presentation to and interaction therewith by a user of the air
traffic demand system 10. In this regard, the demand model 28
represents how many flights will be in various sectors of the
airspace during the time period of interest. The demand model 28 is
updated to incorporate information about the sectors expected to be
traversed by the requested flight 34 and predicted time intervals
that the requested flight 34 is expected to be in such sectors
along with similar information for all other requested flights
analyzed for the time period of interest. In this regard, one or
more additional requested flights (e.g., obtained from the flight
schedule 32) may be analyzed by the air traffic demand prediction
system 10 to generate the demand model 28 for all of the requested
flights during the time period of interest.
FIG. 2 illustrates one embodiment of a case-based retrieval process
200 that may be undertaken by the air traffic demand prediction
system 10 of FIG. 1, and the expanded route prediction component 14
thereof in particular in order to generate the predicted ER.sub.2d
40 associated with the flight request 34. The case based retrieval
process involves querying the historical data 38 for matches using
flight information parameters including the following: (1)
departure airport; (2) destination airport; (3) airline; (4)
aircraft type; (5) flight number; (6) time of day; (7) day of week;
and (8) month of year. If no matches are found using all of the
foregoing parameters, then one or more subsequent queries are
performed until matches are found. Each subsequent query performed
uses progressively fewer parameters (e.g., the first subsequent
query uses parameters (1)-(7), the next subsequent query uses
parameters (1)-(6), etc.).
The matches returned by the query or queries are organized into
clusters based on proximity of geographic position fixes associated
with each flight represented in the historical data 38. The
clusters are created apriori and the matches returned by the query
or queries are sorted according to the historical flight clusters
created apriori. For example, as illustrated in FIG. 2, there may
be a total of eight matches returned that are organized into a
total of four clusters. The first cluster may include three of the
eight matches, the second cluster may include two of the eight
matches, the third cluster may include one of the eight matches,
and the fourth cluster may include two of the eight matches. Thus,
the probabilities associated with the first through fourth clusters
are, respectively, 3/8, 2/8, 1/8, and 2/8. The most represented
cluster (e.g., the first cluster in the example of FIG. 2) is
chosen as the representative cluster and the match with the highest
score (e.g., most matched parameters) is chosen as the seed flight
for the subsequent prediction undertaken by the air traffic demand
prediction system 10.
A cluster selected in accordance with the case-based retrieval
process undertaken by the air traffic demand prediction system 10
may be visualized by plotting rectangular boundaries (bounding
boxes) around geographical position fixes (lat/long points) of the
seed flight. In this regard, FIG. 3 is a plot depicting clustering
of exemplary San Francisco (SFO) to Chicago O'Hare (ORD) routes
that includes four-hundred twenty-four similar flights. In the
example of FIG. 3, bounding boxes that are approximately 1.5
degrees of latitude by 2.5 degrees of longitude have been employed,
but larger or smaller bounding boxes may be employed. The
geographic position fixes (e.g., lat/long points) for the flight
segments located within the bounding boxes surrounding the seed
flight position fixes may be averaged (or otherwise combined in
some manner) to obtain the relevant geometric cluster data.
FIG. 4 illustrates one embodiment of a departure delay prediction
process 400 that may be undertaken by the air traffic demand
prediction system 10 of FIG. 1, and the departure time prediction
component 22 thereof to generate the predicted departure time
information 54 for the requested flight 34. The departure delay
prediction process 400 includes receiving 402 a number of flight
request information parameters including the following: (1)
departure airport; (2) destination airport; (3) airline; (4)
aircraft type; (5) flight number; (6) time of day; (7) day of week;
(8) month of year; and (9) weather conditions at the destination
airport. The flight request information parameters are input to a
case based departure delay module 404. The case based departure
delay module 404 compares the flight request information parameters
input thereto in relation to historical data (e.g., the historical
delay data 52) to identify historically similar cases 406.
The historically similar cases are used to generate a delay
distribution 408. As shown, the delay distribution 408 may be
represented by a curve showing the number of historically similar
cases versus the temporal delay. A predicted delay interval 410 may
then be established. In this regard, the delay interval 410 may be
established using, for example, one standard deviation from the
mean of the distribution.
The delay distribution 408 and predicted delay interval 410 are
input to a departure delay evaluation module 412. The departure
delay evaluation module 412 outputs a temporal prediction interval
414. The temporal prediction interval 414 comprises a predicted
early departure time (earlyStart or ES) and a predicted late
departure time (lateStart or LS) for the requested flight 34. In
this regard, ES may be obtained by subtracting one standard
deviation from the mean departure time of the delay distribution
and LS may be obtained by adding one standard deviation to the mean
departure time of the delay distribution.
FIG. 5A depicts one embodiment of a graph generation process 500
that may be undertaken by the air traffic demand prediction system
10 of FIG. 1, and the graph generation component 26 thereof. The
graph generation process 500 involves propagating relevant
constraints for a plurality of nodes 502A-502D wherein each node
502A-502D represents a sector within the airspace to be traversed
by the requested flight 34. In this regard, the aforementioned
Tachyon software may be utilized to implement the graph generation
process 500 and subsequent solution thereof using applicable
constraints.
In the embodiment of FIG. 5A, there are four nodes 502A-502D, but
there may be more or fewer nodes than depicted. The four nodes
include an initial node 502A, two intermediate nodes 502B, 502C,
and a final node 502D. The initial node 502A represents the first
sector that the requested flight 34 will be in upon entering
controlled airspace (e.g., taking off from the departure airport),
the final node 502D represents the last sector that the requested
flight 34 will be in upon exiting controlled airspace (e.g.,
landing at the destination airport), and the intermediate nodes
502B, 502C represent intermediary sectors entered and exited along
the expected route of the requested flight 34.
A representation of initial node 502A temporal constraints
associated with requested flight 34 is shown in the graph 56 of
FIG. 5B. A number of constraints associated with the requested
flight 34 are depicted in the plot of FIG. 5B, namely an early
start time (ES), a late start time (LS), a minimum elapsed time
(minD) though the sector, and a maximum elapsed time (maxD) through
the sector. The estimated early start (ES) and late start (LS) time
may be obtained in accordance with the departure delay prediction
process 400 as described in connection with FIG. 4. The minD and
maxD constraints may be derived from the predicted sector crossing
information 48 output by the sector crossing component 18 for the
first sector. In addition, an early finish time (EF) and a latest
finish time (LF) depend upon the foregoing constraints (ES, LS,
minD and maxD). As depicted, a total possible time in sector
comprises the difference between LF and ES. Relevant constraints
for the intermediate nodes 502B, 502C and the final node 502D
include minD and maxD for such represented sectors, which may be
derived from the sector crossing information 48 output by the
sector crossing component 18 for such sectors.
The Tachyon temporal reasoner is used to propagate the relevant
constraints for each node 502A-502D to obtain the graph 56
associated with each node 502A-502D. In this regard, FIG. 6 depicts
a solution obtained by the Tachyon temporal reasoner for the four
exemplary sectors represented by the four nodes 502A-502D of FIG.
5A. The solution (shown in the rightmost column of FIG. 6)
represents predicted time intervals during which the requested
flight 34 is expected to be within each of the sectors represented
by the nodes 502A-502D. The predicted time intervals indicate when
the requested flight 34 is expected to be within each sector and
such predicted time intervals are included in the demand model
28.
FIG. 7 depicts one embodiment of a graphical user interface (GUI)
700 of the demand model interface 30 of the air traffic demand
prediction system 10. The GUI includes a number of different panes
or windows 702A-702F. The panes include an airspace information
pane 702A, a sector information pane 702B, a flight information
pane 702C, an events information pane 702D, a control panel pane
702E, and an airspace map pane 702F. The panes 702A-702F may be
arranged in a number of different manners including in a tiled
fashion as depicted.
The airspace information pane 702A displays information identifying
one or more sectors within an airspace and one or more requested
flights within the airspace that have been processed by the air
traffic demand prediction system 10 to include such flights in the
demand model 28. In the example of the FIG. 7, two simulated
requested flights ("EGF264" and "EGF2640") and two sectors ("ZCM06"
and "ZCM25") are listed. During operation of the air traffic demand
prediction system 10 there may be fewer or more requested flights
and fewer or more sectors within the airspace than are listed in
the airspace information pane 702A of the GUI 700 of FIG. 7.
The sector information pane 702B displays information relating to a
selected sector (e.g., selected by clicking on its name in the
airspace information pane 702A or on its location in the airspace
map pane 702F). Information displayed in the sector information
pane 702B may include, for example, total sector load, average
sector load and enroute sector load information. In the example of
FIG. 7, information relating to sector "ZCMO6" is displayed. The
selection of a particular sector for display in the sector
information pane 702B may be indicated by highlighting the selected
sector in the airspace information pane 702A, such as is
illustrated for sector "ZCM06".
The flight information pane 702C displays information relating to a
requested flight processed by the air traffic demand prediction
system 10. Information displayed in the flight information pane
702C may include, for example, flight number, airline, aircraft
type and flight plan (e.g., air speed, cruise level, departure
airport, scheduled departure date/time, destination airport, and
scheduled arrival date/time) information. In the example of FIG. 7,
information relating to flight "EGF264" is displayed since it was
the requested flight most recently processed.
The events information pane 702D displays information relating to
one or more events that may take place for a requested flight
(e.g., the requested flight for which information is displayed in
the flight information pane 702C). In this regard, the information
displayed for each event may include a number of parameters such
as, for example, an event type, the flight identifier (e.g.,
"EGF264"), a sector (e.g., "ZCM25") in which the event occurs, and
the time of the event. Examples of event types include predicted
low (earliest), medium, and high (latest) times of entry of the
flight into a sector and exit of the flight from a sector.
The control panel pane 702E displays information identifying one or
more available air traffic demand predictions (or runs) associated
with one or more airspaces. In the example of FIG. 7, runs
identified as "GBW02", "LIZZI1", "LIZZI2", and "LIZZI3" are
available. A particular run may be selected for execution by the
air traffic demand prediction system 10 by clicking on its
identifier in the control panel pane 702E. In the example of FIG.
7, the selection of the "GBW02" run for execution has been
indicated by highlighting its identifier.
The airspace map pane 702F displays a two-dimensional airspace map
depicting the boundaries of the various sectors within the airspace
associated with the run selected for execution in the control panel
pane 702E. The sector selected for display in the sector
information pane 702B may be highlighted on the map displayed
within the airspace map pane 702F. In the example of FIG. 7, sector
"ZCM06" is highlighted. Additionally, although not shown in FIG. 7,
the various sectors may be color coded to indicate the predicted
sector loads (e.g., total, active, or enroute) associated
therewith. For example, sectors having predicted loads below a
lower acceptable level (e.g., 10 flights) may be color-coded a
first color (e.g., green), sectors having predicted loads between
the lower acceptable level and a higher acceptable level (e.g., 15
flights) may be color coded a second color (e.g., yellow), and
sectors having predicted loads exceeding the higher acceptable
level may be color coded a third color (e.g., red). Such color
coding permits a user of the air traffic demand prediction system
10 to quickly visually identify predicted problem sectors and to
select such sectors for display in the sector information pane
702B. In this regard, a particular sector can also be selected for
display in the sector information pane 702B by selecting it on the
map in the airspace map pane 702F.
While various embodiments of the present invention have been
described in detail, further modifications and adaptations of the
invention may occur to those skilled in the art. However, it is to
be expressly understood that such modifications and adaptations are
within the spirit and scope of the present invention.
* * * * *